丁香园AV

One factor strongly correlating with academic dishonesty is time pressure. In the regular face-to-face (F2F) classroom this can be an issue, and the situation is further exacerbated under conditions such as: force majeure disruptions, large course enrollments, fully online or remote learning delivery, and non-optimal assessment designs.

In a fully online course taught by Dr. Shiv Balram with 240 students completing (245 registered), there were eleven written components including data analysis work that required the evaluation of 2,640 individual student submission during the term. The key question was how can instructors encourage individual students to improve the quality of their submissions, flag potential cases of academic dishonesty, and at the same time maintain a manageable workload for instructors. In response, a semi-automated protocol was implemented outside of the Canvas LMS to flag student submissions for potential academic integrity violations. The design shown in the figure considered three characteristics of the student: course data (submission text and other data), behaviour psychology (interactions), and statistical analysis (clustering and searching).

In a test run, a total of 216 students submissions were processed and the results provided 11 cases forming several groups of potential violations. Each student was contacted and asked for detailed explanations on why their answers would be similar to others. All individuals agreed they indeed worked with members identified in the particular group. This is validation the system is effective in isolating potential cases. The primary reasons students give for the similarities were: (1) worked as a group, (2) used common reference sources, (3) did not paraphrase and cite appropriately, and (4) was unaware collaboration was prohibited. These explanations give further insights into meaningful actions course instructors can take to better help students improve their submissions and avoid direct and indirect academic dishonesty violations.